The overall goal of this study is to identify the dynamic risk and protective processes related to the acquisition of substance use. An integrative developmental contextual model is proposed wherein dynamic processes linking variables across contextual (peer, family, neighborhood and school) and individual (biological, personality, cognitive) systems influence the development of substance use. Through an analysis of interindividual differences and interindividual differences in intraindividual change in substance use, as a function of the age of the child, we will identify those processes that explain why children with similar early risk factors have different developmental trajectories of substance use (multi-finality) and why children with different early risk factors have similar developmental trajectories (equi-finality). Our aims include the description of the development of substance use across a thirteen year span, from 1st grade to one-year post high school, including the identification of clusters of individual trajectories of substance use. We will examine the acquisition of substance use as a function of dynamic processes and key life transitions and will examine the concurrent and across time relations among problem behaviors. Further, we will examine and predict the probability of the occurrence of health-risk outcomes, including unprotected sex, a primary risk factor of HIV-related diseases, as a function of these developmental trajectories of substance use. To accomplish our aims, we propose to continue our current cohort-sequential study wherein over 1000 students in five grade cohorts participated in four annual assessments beginning in the 1st through 5th grade until they were in the 4th through 8th grade. We propose an additional four annual assessments following these participants until they are in the 9th grade through one-year post-high school. Data will continue to be obtained from multiple sources, the child, their parents, teachers and school records allowing the construction of multi-method constructs within the contextual and individual-level systems. Specific hypotheses will be tested using multivariate techniques especially suited for longitudinal data, such as latent growth modeling, latent growth mixture modeling, generalized estimating equations and latent transition analysis. Results will guide the content and timing of substance use prevention and intervention programs.